Personality disorders and suicide attempts in unipolar and bipolar mood disorders

Highlights

• Among mood disorder patients, comorbid personality disorders (PDs) increase the risk of suicide attempts (SAs) to approximately two-fold.

• The excess risk is mostly due to patients with comorbid PDs spending more time in major depressive episodes (MDEs) than those without.

• All DSM-IV PD clusters increased the rate of new SAs, although cluster C PDs more than the others.

• After adjusting for time spent in MDEs, only cluster C further increased the SA rate.

• Direct risk-modifying effects of PDs were also found.

Abstract

Background

Comorbid personality disorders may predispose patients with mood disorders to suicide attempts (SAs), but factors mediating this effect are not well known.

Methods

Altogether 597 patients from three prospective cohort studies (Vantaa Depression Study, Jorvi Bipolar Study, and Vantaa Primary Care Depression Study) were interviewed at baseline, at 18 months, and in VDS and PC-VDS at 5 years. Personality disorders (PDs) at baseline, number of previous SAs, life-charted time spent in major depressive episodes (MDEs), and precise timing of SAs during follow-up were determined and investigated.

Results

Overall, 219 (36.7%) patients had a total of 718 lifetime SAs; 88 (14.7%) patients had 242 SAs during the prospective follow-up. Having any PD diagnosis increased the SA rate, both lifetime and prospectively evaluated, by 90% and 102%, respectively. All PD clusters increased the rate of new SAs, although cluster C PDs more than the others. After adjusting for time spent in MDEs, only cluster C further increased the SA rate (by 52%). Mediation analyses of PD effects on prospectively ascertained SAs indicated significant mediated effects through time at risk in MDEs, but also some direct effects.

Limitations

Findings generalizable only to patients with mood disorders.

Conclusions

Among mood disorder patients, comorbid PDs increase the risk of SAs to approximately two-fold. The excess risk is mostly due to patients with comorbid PDs spending more time in depressive episodes than those without. Consequently, risk appears highest for PDs that most predispose to chronicity and recurrences. However, also direct risk-modifying effects of PDs exist.

Introduction

Mood and personality disorders carry a significant risk of suicide. Between one-half and two-thirds of all suicides occur in patients with mood disorders (Cavanagh et al., 2003), with lifetime risk of suicide in mood disorders in the 5–6% range, the risk being somewhat higher in bipolar disorder (BD) than in major depressive disorder (MDD) (Nordentoft et al., 2011). Moreover, comorbid psychiatric disorders increase suicide risk (Nordentoft et al., 2011). Of individuals who die by suicide, 30–40% of have at least one personality disorder (PD) (Foster et al., 1997, Henriksson et al., 1993) and nearly all of them have also comorbid depressive or substance use disorders, or both (Cheng et al., 1997, Henriksson et al., 1993). Thus, individuals with co-occurring mood and personality disorders comprise the vast majority of all suicides (Foster et al., 1997). Knowing the risk factors for suicide is necessary for rational preventive decisions. However, given suicide’s low base rate, much risk factor research has focused on suicide attempts (SAs) as a proxy for suicide. The risk factors for SAs and suicide are mainly similar, although completers are more often males, have more psychotic symptoms, and use more lethal methods (Hawton et al., 2013, Isometsa, 2014).

Risk factors for SAs in mood disorder patients include previous SAs, younger age, hopelessness, impulsive-aggressive traits, poor perceived social support, and concurrent anxiety, substance use, and PD (for reviews, see (Beghi et al., 2013; Hawton et al., 2013; Isometsa, 2014)). Multiple interacting risk (e.g. above mentioned) and protective (e.g. meaningful and important relationships) factors influence suicidal behavior (Mann et al., 1999). However, in prospective studies (Holma et al., 2010, Holma et al., 2014, Riihimaki et al., 2014a, Valtonen et al., 2008) we have shown that a major determinant, with a consistently high population-attributable fraction, is the time spent in high-risk illness phases. Compared with euthymia, the incidence of SAs during major depressive phases is 25-fold and during mixed illness episodes in BD 65-fold (Holma et al., 2014). Among patients with mood disorders, suicidal acts in the absence of an illness episode are rare (Holma et al., 2010, Holma et al., 2014, Riihimaki et al., 2014a, Valtonen et al., 2008). Both comorbid DSM-IV cluster B (Skodol et al., 2011) and C (Oleski et al., 2012) PDs have been shown to increase time that an individual spends in major depressive episodes (MDEs).

The relationship between mood disorder and comorbid PDs and SAs has been investigated in longitudinal epidemiological (Bolton et al., 2010, Oleski et al., 2012) and clinical (Galfalvy et al., 2006, Oquendo et al., 2007, Stringer et al., 2013, Undurraga et al., 2012) studies among mood disorder patients and in longitudinal clinical studies among PD patients (Soloff and Chiappetta, 2012, Yen et al., 2003). In these studies, cluster B PDs, especially borderline PD, have been established as a risk factor for SAs, whereas cluster A and C PDs have received less attention. The most consistent predictors for a future SA have been borderline PD comorbid with MDD (Bolton et al., 2010, Soloff and Chiappetta, 2012, Soloff and Fabio, 2008, Yen et al., 2003), BD with current MDE (Galfalvy et al., 2006) , and both MDD and BD with current MDE (Oquendo et al., 2007) or dysthymia (Stringer et al., 2013). Also baseline cluster C (Oleski et al., 2012), dependent (Bolton et al., 2010), avoidant (Bolton et al., 2010), paranoid (Bolton et al., 2010), schizoid (Bolton et al., 2010), schizotypal (Bolton et al., 2010), or any (Bolton et al., 2010, Undurraga et al., 2012) PD comorbid with MDD or MDE in BD patients has been associated with higher rates of SAs at follow-up. A number of other risk factors e.g. sociodemographic factors, number of depressive symptoms, other comorbid disorder, previous suicidal behavior have been controlled in the aforementioned studies, but not the proportion of time spent in MDEs during the follow-up; yet, it is a key predictor of accumulated risk for suicidal acts (Holma et al., 2010, Holma et al., 2014, Riihimaki et al., 2014a, Valtonen et al., 2008).

In this prospective cohort study of mood disorder patients, we aimed to investigate the relationship between comorbid PDs and future SAs. We hypothesized that 1) baseline comorbid cluster C and borderline PDs would increase the rate of SAs indirectly by increasing time spent in MDEs. Moreover, 2) we expected that borderline PD would also (directly) modify the risk of SAs during MDEs.

Therefore, we investigated the relationship between comorbid PD and 1) lifetime (total retrospective and prospective) SAs and 2) prospectively evaluated new SAs during the follow-up. We also tested 3) whether comorbid PD acts directly on the rate of SAs or indirectly by increasing the duration of MDEs during the follow-up. In analyses of prospectively evaluated SAs, we 4) controlled for confounding sociodemographic and clinical risk factors. Finally, we conducted formal mediation analyses.